Search for: All records

Objectives: Early identification of sepsis is critical, as delayed diagnosis significantly increases morbidity and mortality. We aimed to develop and validate AI/ML models for early sepsis prediction using structured electronic health record (EHR) data, waveform data, and a combination of both. Methods: We conducted a retrospective observational study using the AIM-AHEAD60 subset of the CHoRUS dataset. Adult patients (≥18 years) with a final diagnosis of sepsis were included. Structured EHR data (demographics, initial vital signs, laboratory results) and waveform data (continuous vital signs) from the first hour of hospital arrival were extracted. Three algorithms (i.e., XGBoost, LightGBM, and HistGB) were developed with a focus on maximizing the performance metric of recall. Other performance metrics were also assessed, including accuracy, precision, F1 score, and the area under the receiver operating characteristic curve (AUROC). Results: A total of 11,312 unique patients met the inclusion criteria, among whom 2245 individuals (19.85%) were diagnosed with sepsis at least once. Using structured EHR data alone, laboratory variables such as lactate and leukocyte count were most predictive. Waveform models identified respiratory rate, systolic blood pressure, and temperature trends in the first hour as key predictors. Combined models highlighted mean temperature and mean systolic blood pressure as top features. XGBoost achieved the highest AUROC (0.922) across all data configurations, with a recall above 80%, demonstrating robust performance despite substantial missing data. Conclusions: High-performing AI/ML models for early sepsis prediction can be developed from publicly available datasets using only first-hour clinical information. XGBoost models demonstrate strong potential for real-time clinical screening. 

Methyl 2-acetamido-2-deoxy-β-D-glucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-D-glucopyranoside (methyl β-chitobioside), (IV), crystallizes from aqueous methanol at room temperature to give a structure (C₁₇H₃₀N₂O₂₂·CH₃OH) containing conformational disorder in the exocyclic hydroxymethyl group of one of its βGlcNAc residues. As observed in other X-ray structures of disaccharides containing β-(1→4)O-glycosidic linkages, inter-residue hydrogen bonding between O3H of the βGlcNAc bearing the OCH₃aglycone and O5 of the adjacent βGlcNAc is observed based on the 2.79 Å internuclear distance between the O atoms. The structure of (IV) was compared to that determined previously for 2-acetamido-2-deoxy-β-D-glucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-D-glucopyranose (β-chitobiose), (III). TheO-glycosidic linkage torsion angles,phi(ϕ) andpsi(ψ), in (III) and (IV) differ by 6–8°. TheN-acetyl side chain conformation in (III) and (IV) shows some context dependence, with the C1—C2—N—C_cartorsion angle 10–15° smaller for the βGlcNAc residue involved in the internalO-glycosidic linkage. In (IV), conformational disorder is observed in the exocyclic hydroxymethyl (–CH₂OH) group in the βGlcNAc residue bearing the OCH₃aglycone, and a fitting of the electron density indicates an approximate 50:50 distribution of thegauche–gauche(gg) andgauche–trans(gt) conformers in the lattice. Similar behavior is not observed in (III), presumably due to the different packing structure in the vicinity of the –CH₂OH substituent that affects its ability to hydrogen bond to proximal donors/acceptors. Unlike (IV), a re-examination of the previously reported electron density of (III) revealed conformational disorder in theN-acetyl side chain attached to the reducing-end βGlcNAc residue caused by rotation about the C2—N bond. 

Two disaccharides, methyl β-d-galactopyranosyl-(1→4)-α-d-glucopyranoside (1) and methyl β-d-galactopyranosyl-(1→4)-3-deoxy-α-d-ribo-hexopyranoside (3), were prepared with selective 13C-enrichment to allow measurement of six trans-O-glycosidic J-couplings (2JCOC, 3JCOCH, and 3JCOCC) in each compound. Density functional theory (DFT) was used to parameterize Karplus-like equations that relate these J-couplings to either ϕ or ψ. MA’AT analysis was applied to both linkages to determine mean values of ϕ and ψ in each disaccharide and their associated circular standard deviations (CSDs). Results show that deoxygenation at C3 of 1 has little effect on both the mean values and librational motions of the linkage torsion angles. This finding implies that, if inter-residue hydrogen bonding between O3H and O5′ of 1 is present in aqueous solution and persistent, it plays little if any role in dictating preferred linkage conformation. Hydrogen bonding may lower the energy of the preferred linkage geometry but does not determine it to any appreciable extent. Aqueous 1-μs MD simulation supports this conclusion and also indicates greater conformational flexibility in deoxydisaccharide 3 in terms of sampling several, conformationally distinct, higher-energy conformers in solution. The populations of these latter conformers are low (3–14%) and could not be validated by MA’AT analysis. If the MD model is correct, however, C3 deoxygenation does enable conformational sampling over a wider range of ϕ/ψ values, but linkage conformation in the predominant conformer is essentially identical in both 1 and 3. 

The emergence of Intel's Optane DC persistent memory (Optane Pmem) draws much interest in building persistent key-value (KV) stores to take advantage of its high throughput and low latency. A major challenge in the efforts stems from the fact that Optane Pmem is essentially a hybrid storage device with two distinct properties. On one hand, it is a high-speed byte-addressable device similar to DRAM. On the other hand, the write to the Optane media is conducted at the unit of 256 bytes, much like a block storage device. Existing KV store designs for persistent memory do not take into account of the latter property, leading to high write amplification and constraining both write and read throughput. In the meantime, a direct re-use of a KV store design intended for block devices, such as LSM-based ones, would cause much higher read latency due to the former property. In this paper, we propose ChameleonDB, a KV store design specifically for this important hybrid memory/storage device by considering and exploiting these two properties in one design. It uses LSM tree structure to efficiently admit writes with low write amplification. It uses an in-DRAM hash table to bypass LSM-tree's multiple levels for fast reads. In the meantime, ChameleonDB may choose to opportunistically maintain the LSM multi-level structure in the background to achieve short recovery time after a system crash. ChameleonDB's hybrid structure is designed to be able to absorb sudden bursts of a write workload, which helps avoid long-tail read latency. Our experiment results show that ChameleonDB improves write throughput by 3.3× and reduces read latency by around 60% compared with a legacy LSM-tree based KV store design. ChameleonDB provides performance competitive even with KV stores using fully in-DRAM index by using much less DRAM space. Compared with CCEH, a persistent hash table design, ChameleonDB provides 6.4× higher write throughput. 

The crystal structure of methyl 2-acetamido-2-deoxy-β-D-glycopyranosyl-(1→4)-β-D-mannopyranoside monohydrate, C 15 H 27 NO 11 ·H 2 O, was determined and its structural properties compared to those in a set of mono- and disaccharides bearing N -acetyl side-chains in βGlcNAc aldohexopyranosyl rings. Valence bond angles and torsion angles in these side chains are relatively uniform, but C—N (amide) and C—O (carbonyl) bond lengths depend on the state of hydrogen bonding to the carbonyl O atom and N—H hydrogen. Relative to N -acetyl side chains devoid of hydrogen bonding, those in which the carbonyl O atom serves as a hydrogen-bond acceptor display elongated C—O and shortened C—N bonds. This behavior is reproduced by density functional theory (DFT) calculations, indicating that the relative contributions of amide resonance forms to experimental C—N and C—O bond lengths depend on the solvation state, leading to expectations that activation barriers to amide cis – trans isomerization will depend on the polarity of the environment. DFT calculations also revealed useful predictive information on the dependencies of inter-residue hydrogen bonding and some bond angles in or proximal to β-(1→4) O -glycosidic linkages on linkage torsion angles ϕ and ψ. Hypersurfaces correlating ϕ and ψ with the linkage C—O—C bond angle and total energy are sufficiently similar to render the former a proxy of the latter.